arxiv:astro-ph/0603037v1 2 mar 2006se direction (135 ) but this has not yet been measured. there is...

arX

iv:a

stro

-ph/

0603

037v

1 2

Mar

200

6Mon. Not. R. Astron. Soc. 000, 1–16 (2006) Printed 20 August 2018 (MN LATEX style file v1.4)

Profile morphology and polarization of young pulsars

Simon Johnston1 and Joel M. Weisberg21Australia Telescope National Facility, CSIRO, P.O. Box 76, Epping, NSW 1710, Australia.2Dept. of Physics and Astronomy, Carleton College, Northfield, MN 55057, USA.

20 August 2018

ABSTRACTWe present polarization profiles at 1.4 and 3.1 GHz for 14 young pulsars with char-acteristic ages less than 75 kyr. Careful calibration ensures that the absolute positionangle of the linearly polarized radiation at the pulsar is obtained. In combination withpreviously published data we draw three main conclusions about the pulse profiles ofyoung pulsars. (1) Pulse profiles are simple and consist of either one or two promi-nent components. (2) The linearly polarized fraction is nearly always in excess of 70per cent. (3) In profiles with two components the trailing component nearly alwaysdominates, only the trailing component shows circular polarization and the positionangle swing is generally flat across the leading component and steep across the trailingcomponent.

Based on these results we can make the following generalisations about the emis-sion beams of young pulsars. (1) There is a single, relatively wide cone of emissionfrom near the last open field lines. (2) Core emission is absent or rather weak. (3) Theheight of the emission is between 1 and 10 per cent of the light cylinder radius.

Key words: pulsars:general

1 INTRODUCTION

The polarization profiles of radio pulsars have long been avaluable tool for a variety of different purposes. They can beused, for example, to classify the different types of profile,to understand the underlying emission mechanism and todetermine the geometry of the star. In the rotating vectormodel (RVM) of Radhakrishnan & Cooke (1969), the radi-ation is beamed along the field lines and the plane of polar-ization is determined by the angle of the magnetic field asit sweeps past the line of sight. The position angle (PA) asa function of pulse longitude, φ, can be expressed as

PA = PA0 + arctan

(

sinα sin(φ− φ0)

sinζ cosα− cosζ sinα cos(φ− φ0)

)

(1)

Here, α is the angle between the rotation axis and the mag-netic axis and ζ = α + β with β being the angle of closestapproach of the line of sight to the magnetic axis. φ0 is thepulse longitude at which the PA is PA0, which also corre-sponds to the PA of the rotation axis projected onto theplane of the sky. We note that RVM fitting does not de-pend on the total intensity profile, or this location of theprofile symmetry points. Unfortunately, for most pulsars itis difficult to determine α and β with any degree of accu-racy, partly because the longitude over which pulsars emitis rather small and partly because strong deviations froma simple swing of PA are often observed. This makes thedetermination of the various angles straightforward only in

the ∼15 per cent of pulsars for which the RVM works (seethe discussion in Everett & Weisberg 2001).

If the emission occurs at some height above the pulsarsurface, the PA swing can be delayed with respect to the to-tal intensity profile by relativistic effects such as aberrationand retardation (hereafter referred to as A/R) as initiallydiscussed by Blaskiewicz, Cordes & Wasserman (1991). Themagnitude of the shift in longitude, δφ(PA) (in radians), isrelated to the emission height relative to the centre of thestar, hem, via

δφ(PA) =8 π hem

P c(2)

where P is the pulsar period and c the speed of light (Blask-iewicz et al. 1991). We note that, at least to first order, thePA swing is not altered by A/R effects other than the longi-tude delay. These effects therefore do not affect the measuredvalues of α and β from equation 1.

There is also a geometrical method which can be used tocompute emission heights if the angles α and β are known.Under the assumption of a dipolar field and a circular emis-sion zone, the half-opening angle of the emission cone, ρecan be expressed as

cosρe = cosα cosζ + sinα sinζ cos(W/2) (3)

where W is the measured pulse width in longitude (Gil,Gronkowski & Rudnicki 1984). If one assumes that the emis-

c© 2006 RAS

http://arxiv.org/abs/astro-ph/0603037v1

2 Johnston & Weisberg

sion extends to the final open field line then the emissionheight can be derived through the expression

ρo = 3

√

π hem2 P c

(4)

(Rankin 1990). Here we use ρo to denote the half-openingangle of the cone at the last open field line, and note thatunless the emission extends right to the edge of the conethen ρo > ρe.

It is possible to compute relative emission heights ofthe radiation without the polarization information (Gan-gadhara & Gupta 2001). Imagine the cone emission comesfrom a circular symmetric region around the core. Then, ifcone emission arises higher in the magnetosphere than thecore emission, the entire circular structure will be shiftedrelative to the core due to A/R effects and the profile willappear asymmetric. If a given profile contains well definedcore and conal emission then the difference in separationbetween the leading and trailing conal components and thecore, δφ(CC), can be used to compute their relative emissionheights, ∆hem. In this case

δφ(CC) =4 π ∆hem

P c(5)

(Dyks, Rudak & Harding 2004). Gangadhara (2005) hasmodified this equation to take into account the viewing ge-ometry which can affect the derived emission heights by∼10 per cent in some cases. It can be useful to comparethe emission height with the radius of the light cylinder,rlc = Pc/2π. Generally it is found that the radio emissionoccurs in regions significantly below about 0.1 rlc. Substan-tial discussion of the merits and failings of all these methodscan be found in recent papers by e.g. Gupta & Gangadhara(2003), Mitra & Li (2004), Dyks et al. (2004) and Gangad-hara (2005).

Young pulsars and/or pulsars with high spin down en-ergies, as a group, tend to have high (> 70 per cent) lin-ear polarization (e.g. Qiao et al. 1995, Crawford et al. 2001)and often show simple Gaussian-like profiles with little struc-ture. Some however, such as PSRs B1259−63 (Manchester &Johnston 1995) and B0906−49 (Qiao et al. 1995) have widedouble profiles. Also, as a class, young pulsars tend to haveflatter spectral indices than the older population. Many haveassociated non-thermal high energy emission whose profilesare offset from the radio profiles. These features led Manch-ester (1996) to suggest that emission from young pulsarsoccurred relatively near the light cylinder.

Han et al. (1998) showed a remarkable correlation be-tween the circular polarization properties of pulsars and thedirection of their position angle swing. For pulsars whichshow simple (conal) double profiles, the sign of circular po-larization under the components is correlated with the slopeof the PA swing; left-hand circular polarization implies anegative slope and vice-versa. There appear to be no excep-tions to this rule in a sample which has now grown to 40pulsars (You & Han 2005). However, this correlation doesnot hold for other pulsars with more complicated pulse pro-files.

The supernova explosion that creates the pulsar alsoproduces a supernova remnant (SNR) and many young pul-sars are seen, at least in projection, inside SNR shells.One way to confirm the association between the pulsar

and the SNR is to obtain the proper motion for the pul-sar and determine whether it originated from the SNR cen-tre. Only a relatively small number of young pulsars havegood proper motion measurements but polarization obser-vations also provide valuable information. Johnston et al.(2005) have recently used the polarization from young pul-sars to show that, for the majority of cases, the rotationaxis is aligned with the velocity vector. Therefore, for pul-sars without known proper motions, it may be possible toobtain the direction of motion directly from the directionof the rotation axis inferred through polarization measure-ments.

The known pulsars can be sorted according to theircharacteristic ages (τc = P/2Ṗ , where P is the pulsar spinperiod and Ṗ is the period derivative) or their spin downenergy, Ė (proportional to Ṗ /P 3). Naturally, the youngestpulsars tend also to have the highest Ė. One of the aims ofthis paper is to determine whether, as a class, the young,highly energetic pulsars have characteristic total intensityand polarization profiles.

A significant number of young pulsars already havemeasured polarization properties by a variety of authors overa range of frequencies albeit with rather low time resolutionand without absolute PA determination. However, recentpulsar surveys have uncovered many more young objects forwhich the total intensity profiles are available but with fewpolarization measurements. We selected 10 of these pulsarsto observe at both 1.4 and 3.1 GHz, and a further 4 whichwere observed at 1.4 GHz only. The selection was based onthe following criteria: (1) Characteristic ages of less than 75kyr, (2) Right Ascension less than 17 h, (2) Declinations lessthan 0◦. Many of the pulsars also have a possible associationwith high energy emission or with an SNR.

The outline of the paper is as follows. In Section 2 wedescribe in more detail those pulsars which have associationswith high energy emission and/or SNRs. In Section 3 we de-scribe the observations and present the results in Section 4,with Section 5 detailing fits of the rotating vector model. InSection 6 we discuss the polarization of young pulsars gen-erally, and in Section 7 we analyse the implications of thepolarization results for the proper motion direction.

2 PULSARS WITH POSSIBLE HIGH ENERGYOR SNR ASSOCIATIONS

PSR J1015−5719 was discovered by Kramer et al. (2003)and they and Torres et al. (2001) pointed out thatit is located near the unidentified EGRET source 3EGJ1014−5705. The energetics make the association plausibleif the distance to the pulsar is ∼5 kpc.

PSR J1016−5857 was discovered by Camilo et al.(2001). It is located in projection near SNR G284.3−1.8 andalthough the distance to the SNR and to the pulsar disagree,Camilo et al. (2001) consider the association to be likely. Thepulsar is located about 15 arcmin from the apparent SNRcentre in a WNW direction. If the association is correct, theproper motion direction should be ∼300◦. The pulsar maybe associated with the EGRET source 3EG J1013−5915.

PSR J1105−6107 was discovered by Kaspi et al. (1997).It is a young pulsar and is not obviously embedded in a par-ent SNR. However, SNR G290.1−0.8 is nearby (in projec-

c© 2006 RAS, MNRAS 000, 1–16

Profile morphology and polarization of young pulsars 3

tion) and Kaspi et al. (1997) discuss the possibility that thetwo are associated. If so, its proper motion should be in aSE direction (135◦) but this has not yet been measured.There is a possible association with the EGRET source2EG J1103−6106. Crawford et al. (2001) provide a polar-ization profile of this pulsar at 1.4 GHz and computed itsRM to be 166 rad m−2.

PSR J1119−6127 was discovered by Camilo et al. (2000)and its environs have been imaged in the radio (Crawfordet al. 2001) and the X-ray (Pivovaroff et al. 2001). It liesvirtually in the centre of the shell-like SNR G292.2−0.5 andthe association appears secure. Gonzalez et al. (2005) haverecently detected a pulsar wind nebula and thermal pulsa-tions from the pulsar in the X-ray. Crawford & Keim (2003)published polarization results for this pulsar at 1.4 GHz andshowed that it had a high degree of linear polarization andan RM of 842 rad m−2.

PSR J1341−6220 (B1338−62) was discovered byManchester et al. (1985). Timing of the pulsar provedit to be young (Kaspi et al. 1992) and a high resolu-tion image by Caswell et al. (1992) revealed the pul-sar to be located near the centre of SNR G308.8−0.1.One might then expect the proper motion direction to be∼315◦. The pulsar/SNR system is rather similar to thePSR B1509−58/SNR G320.4−1.2 association discussed inmore detail below. Qiao et al. (1995) and Crawford et al.(2001) provide polarization profiles of this pulsar at 1.4 GHzwhich show a profile largely dominated by scattering. TheRM of the pulsar is −946 rad m−2.

PSRs J1412−6145 (Manchester et al. 2001) andJ1413−6141 (Kramer et al. 2003) were discovered as partof the Parkes multibeam survey. Both pulsars lie within theSNR G312.4−0.4 though it is not clear which, if either, isassociated with it (Doherty et al. 2003). Proper motionsfor these pulsar would help resolve this issue. The errorbox of the EGRET source 3EG J1410−6147 encompassesthe SNR and the two pulsars. Both Case & Bhattacharya(1999) and Doherty et al. (2003) discuss the possibility thatthe EGRET source and the SNR and/or the pulsar(s) areassociated without reaching any firm conclusions.

PSRs J1420−6048 was discovered by D’Amico et al.(2001). Prior to the detection of the pulsar, Roberts etal. (1999) made a radio image of the area surrounding theEGRET source 2EG J1418−6049 and detected a wind neb-ula (the ‘Rabbit’) surrounded by a possible non-thermalshell (the ‘Kookaburra’). They concluded that a young en-ergetic pulsar was likely the source of both the wind nebulaand the EGRET source. However, the discovered pulsar liessomewhat outside the wind nebula, further muddying an al-ready complex picture. X-ray pulsations have been detectedfrom the pulsar (Roberts et al. 2001) and it remains a can-didate for the EGRET source. Roberts et al. (2001) alsopresent the polarization profile of the pulsar at 1.4 GHz andderive an RM of −106 rad m−2.

PSR J1513−5908 (B1509−58) was first discovered inX-rays (Seward et al. 1982) and subsequently in the radio(Manchester et al. 1982) and is likely associated and inter-acting with SNR G320.4−1.2. In an exhaustive radio studyof the complex, Gaensler et al. (1999) make two commentsof relevance here. First they speculate that the star Muzzio10 may have been a (former) companion of the pulsar andhence the pulsar’s proper motion should be at a position

angle of 168◦. From morphology considerations they arguethat the pulsar’s rotation axis must point at an angle of ap-proximately 145◦ and 315◦ (see also Brazier & Becker 1997).Observations of the wind nebula around the pulsar in the X-ray, showed that the nebula has a clear symmetry axis withposition angle of 150◦ (Gaensler et al. 2002), consistent withthe radio observations. Polarization profiles for this pulsarwere shown at 0.66 and 1.4 GHz by Crawford et al. (2001).At 0.66 GHz the pulse is extremely scatter broadened butat 1.4 GHz shows a simple Gaussian profile extending overabout 70◦ of longitude.

3 OBSERVATIONS

Observations were carried out using the 64-m radio tele-scope located near Parkes, New South Wales. Two differ-ent receiver packages were used at frequencies near 1.4 and3.1 GHz. The H-OH receiver covers the frequency range from1.2 to 1.7 GHz. We used a central frequency of 1.368 GHzand a total bandwidth of 256 MHz. The 10cm part of thedual frequency 10/50 cm receiver has a total bandwidth of1024 MHz centered at 3.1 GHz. All the observations tookplace in 2005. The majority of the 1.4 GHz observationswere made in the period July 9 to 13 and the 3.1 GHz ob-servations made between June 20 and 26.

The pulsars were typically observed for 30 mins each,preceded by a 2 min observation of the pulsed calibrator.The total bandwidth was subdivided into 1024 frequencychannels and the pulsar period divided into 1024 phase binsby the backend correlator. The correlator folds the data for60 s at the topocentric period of the pulsar at that epochand records the data to disk for offline-processing. Duringthe observing session, observations were made of the fluxcalibrator Hydra A whose flux density is 43 and 21 Jy at theobserving frequencies of 1.4 and 3.1 GHz. This allowed usto determine the system equivalent flux density to be 28.7and 46.2 Jy and allowed for flux calibration of the pulsarprofiles.

The data were analysed off-line using the PSRCHIVEsoftware package (Hotan et al. 2005). Polarization calibra-tion was carried out using the observations of the pulsedcalibrator signal to determine the relative gain and phasebetween the two feed probes. The data were corrected forparallactic angle and the orientation of the feed. The po-sition angles were also corrected for the Faraday rotationalong the path to the pulsar (i.e. the ionospheric and inter-stellar medium contributions) using the rotation measure(RM) determined from the data. Hence, all PAs are abso-lute PAs at the pulsar and can thus be compared directly atdifferent frequencies. Flux calibration was carried out usingthe Hydra A observations. The final product was thereforeflux and polarization calibrated Stokes I , Q, U , V profiles.

The contributions to the measured values of RM aris-ing in the ionosphere have been estimated by integratingthe International Reference Ionosphere and InternationalGeomagnetic Reference Field time-dependent models of theionospheric electron density and geomagnetic field throughthe ionosphere along the sight line between the telescope andthe pulsar (Bilitza 2003) using an implementation providedby JL Han. Values of the ionospheric RM range from 0 to

c© 2006 RAS, MNRAS 000, 1–16


Table 1. Pulsar parameters

Jname Bname Period Age T1.4/T3.1 N1.4/N3.1 RM S1.4 S3.1 W101.4 W103.1(ms) (kyr) (s) rad m−2 (mJy) (mJy) (deg) (deg)

J0729−1448 251.7 35.2 5400/6000 256/1024 46±1 0.50 0.43 25 24J0940−5048 87.5 42.2 3600/6000 256/512 −31±1 0.65 0.47 43 42J1015−5719 139.9 38.6 1800 256 96±2 3.50 155J1016−5857 107.4 21.0 3600/6000 256/512 −540±3 0.81 0.38 36 31J1105−6107 63.2 63.6 1800/3600 256/512 187±1 1.25 0.63 23 25J1119−6127 407.8 1.6 3600/3600 256/256 853±2 0.99 0.44 43 34J1301−6305 184.5 11.0 1800 256 −625±25 1.00 80J1341−6220 B1338−62 193.3 12.1 1800/1200 256/512 −921±3 2.75 42 15J1357−6429 166.1 7.3 1800 256 −47±2 45J1412−6145 315.2 50.6 3600/1200 256/256 −130±13 0.64 0.24 21 18J1413−6141 285.6 13.6 5400/2400 256/256 −400±40 0.85 0.54 72 7J1420−6048 68.2 13.0 5400/2400 256/512 −122±5 1.20 73 58J1513−5908 B1509−58 150.7 1.6 5400/5400 256/128 216±1 1.66 0.46 103 127J1702−4310 240.5 17.0 5400 256 −35±12 1.12 35

−2 rad m−2 and were subtracted from the measured valuesto leave only the interstellar components.

4 RESULTS

Table 1 lists the pulsars observed along with their period(column 3) and characteristic age (column 4). Column 5displays the total integration time at 1.4 and 3.1 GHz andcolumn 6 gives the number of phase bins in the profiles atthe two frequencies. Column 7 gives the RM due to the inter-stellar medium alone, columns 8 and 9 list the flux densitiesat the two observing frequencies. The final two columns givethe width of the profile at the 10 per cent intensity level.

Figures 1-6 show the calibrated polarization profiles forthe pulsars in our sample. We have located zero longitudeat the symmetrical midpoint given by the 10 per cent in-tensity level of the profiles in most cases. The exceptionsto this rule are the 1.4 GHz profiles of PSRs J1341−6220and J1413−5141 which are scatter broadened; the peak ofthe profile was aligned with that at 3.1 GHz. Results forindividual pulsars are described in detail below.PSR J0729−1448: The profile of this pulsar undergoessignificant frequency evolution between 1.4 and 3.1 GHz. Atboth frequencies there are 3 clearly visible components; thisis one of only two pulsars in our sample to show a centralcomponent. At the lower frequency, the trailing componentdominates and the leading component is weakest, whereasat the higher frequency the leading and trailing componentshave roughly equal strength. The linear polarization at bothfrequencies is high throughout with positive circular polar-ization seen against the trailing component. The PA swingsteepens sharply through the trailing component. In spiteof the strong frequency evolution, the profiles widths arevirtually identical at both frequencies.PSR J0940−5428: The pulse profile is very similar at both1.4 and 3.1 GHz showing what appears to be a classic dou-ble structure with a stronger trailing component. The saddleregion is somewhat better filled in at the higher frequencyeven though the overall pulse widths are similar at both fre-quencies. The linear polarization is high at both frequenciesand there is a hint of circular polarization against the trail-

ing component. The PA swing is steeper against the secondcomponent.

PSR J1015−5719: Observations were made only at1.4 GHz for this pulsar. The profile is a wide double withemission over nearly 200 degrees of longitude. A small cen-tral component is also visible. It is interesting that the steepedges of the profile appear on the interior rather than theexterior of the components. This is the opposite to what isgenerally observed. Circular polarization is present againstboth the leading and trailing component. The linear polar-ization is similar against both components; it is very highexcept for the wings of the pulse where it is virtually zero.This implies that both components must themselves be com-posed of several individual components. We will return tothis point later. It appears likely that there is an orthogonaljump between the central and trailing components.

PSR J1016−5857: Similar to PSR J0940−5428, the pro-file shows two components of which the trailing componentis dominant. The polarization is low in the leading com-ponent but relatively high in the trailing component, as isthe circular polarization. The behaviour of the PA acrossthe leading component is unclear. Across the trailing com-ponent, the PA seems rather flat at 1.4 GHz but appearsto have significant negative slope at 3.1 GHz. The profile iswider at the lower freqeuency.

PSR J1105−6107: The profile at 1.4 GHz is similar tothat seen in Crawford et al. (2001) except that our bettertime resolution clearly splits the two components. The pulseprofile consists of two components, nearly equal in strengthat 1.4 GHz but with the trailing component dominant at3.1 GHz. The linear polarization is high throughout but cir-cular polarization is only present under the trailing compo-nent. The PA swing is flat under the leading component andsteepens under the trailing component.

PSR J1119−6127: The pulse profile is a similar quasi-Gaussian shape at both frequencies although the 3.1 GHzprofile is slightly narrower than its 1.4 GHz counterpart. Thepercentage linear polarization is high throughout the profieand the PA swing is flat.

PSR J1301−6305: The profile consists of a single com-ponent which occupies about 100 degrees of longitude. Itis highly polarized and shows a shallow swing of position

c© 2006 RAS, MNRAS 000, 1–16


Figure 1. Polarization profiles at 1.4 GHz and 3.1 GHz for PSRs J0729−1449 and J0940−5428. The top panel of each plot shows thePA variation with respect to celestial north as a function of longitude. The PAs are corrected for RM and represent the (frequencyindependent) value at the pulsar. The lower panel shows the integrated profile in total intensity (thick line), linear polarization (darkgrey line) and circular polarization (light grey line). The choice of longitude zero for each pulsar at the symmetrical midpoint of theprofile and the profiles at the two frequencies have been aligned as closely as possible.

c© 2006 RAS, MNRAS 000, 1–16


Figure 2. Polarization profiles at 1.4 GHz and 3.1 GHz for PSRs J1016−5857 and J1105−6107. See Fig 1 for details.

c© 2006 RAS, MNRAS 000, 1–16


Figure 6. Polarization profiles for PSRs J1015−5719, J1301−6305, J1357−6429 amd J1702−4310 which were observed only at 1.4 GHz.See Fig 1 for details.

c© 2006 RAS, MNRAS 000, 1–16


Table 2. Constraints on geometric angles and emission heights from RVM fits

Jname Bname rlc α |β| φ0 PA0 hem hem/rlc ρo(km) (deg) (deg) (deg) (deg) (km) (deg)

J0729−1448 12000 –


Figure 7. RVM fitting to PA swing in PSR J1015−5719. The left hand panel shows the χ2 contours in the α-β plane. The right handpanel shows the data (with the addition of an orthogonal jump at longitude ∼15◦, cf Fig. 6) and the best fit with α=101◦, β=−20◦.

an RVM fit yields α ∼ 86◦, β ∼ −10◦ and the best fit linepasses underneath the majority of the central component PAvalues. In this case, φ0 lies at a longitude of ∼250

◦, signifi-cantly after the emission locations. Whichever fit is chosen,small values of α are ruled out and the pulsar appears to bean orthogonal rotator viewed at a rather large impact an-gle. Note that the α and β values quoted here are correctedfor the ‘PA convention problem’ (Everett & Weisberg 2001),wherein observed PAs increase counter-clockwise while theRVM model assumes the opposite convention.

For PSR J1119−6127, we get similar results to Craw-ford & Keim (2003) in that β must be less than about20◦, and α is almost unconstrained. As they do, we alsoderive φ0 to be on the far trailing edge of the pulse pro-file. Although there is no linear polarization at this loca-tion we can estimate PA0 to be ∼33

◦from the RVM fit. ForPSR J1420−6048, we find that α and β cannot be as wellconstrained as claimed by Roberts et al. (2001) who derivedα < 35◦ and β ∼ 0.5◦. For PSR J1513−5908, contrary toCrawford et al. (2001), we cannot constrain α, and we there-fore cannot exclude the estimate of 60◦ made by Romani &Yadigaroglu (1995) from the high energy pulse profile. Inter-pretation of the X-ray morphology of the pulsar wind neb-ula yields evidence that α+ β is greater then 70◦(Brazier &Becker 1997; Yatsu et al. 2005). If this value is correct, ourfit then constrains α to lie between 30◦ and 150◦.

6 POLARIZATION OF YOUNG PULSARS

It is evident even from a cursory glance at Figs 1−6 thatmany of the profiles of these young pulsars have remarkablysimilar characteristics. First, we see that these pulse pro-files are relatively simple, unlike the highly complex profilesseen in older pulsars (as already pointed out decades agoby Huguenin et al. 1971). Lyne & Manchester (1988) con-

cluded that this was because the relative spectral index ofcore and cone component was age dependent. Of the 14 pul-sars studied, 9 are double profiled with the trailing compo-nent dominating in all cases. Some have almost equal ampli-tude components (e.g. PSR J1420−6048) but in others theratio is rather large (e.g. PSR J1513−5908). Seven of the9 which have measurable circular polarization have so onlyunder the trailing component. All 9 show a flat PA swing fol-lowed by a steeper swing under the trailing component. Thefive pulsars which do not have multi-component profiles arePSRs J1119−6127, J1301−6305, J1357−6492, J1412−6145and J1702−4310. These have single components and a ratherflat PA swing with little or no circular polarization. Finally,12 of the 14 have large degrees of linear polarization (in ex-cess of 50 per cent) and no obvious orthogonal mode jumpsin the profiles with the exception being PSR J1015−5719and possibly PSR J1413−6141. The spectral index evolu-tion between these two frequencies shows no obvious trendsin the 8 pulsars with two components and dual frequencymeasurements. Three show a flatter index leading compo-nent, three have a flatter index trailing component and theremaining two have similar indicies for both leading andtrailing components!

We note that these features are not only seen in thepresent data but also in previous polarization observationsof young pulsars (Qiao et al. 1995; von Hoensbroech, Ki-jak & Krawczyk 1998; Crawford, Manchester & Kaspi 2001;Karastergiou, Johnston & Manchester 2005). Nearly all havehigh linear polarization and those pulsars with two com-ponents have the trailing component dominant (e.g. PSRsB1610−50, B1643−43, B1758−23, B1800−21, B1823−13).The only exception to this rule is the Vela pulsar in whichthe leading component dominates the profile below 5 GHz.

There are two possible explanations for the morphologyof these young pulsars. In the first scenario, the two observed

c© 2006 RAS, MNRAS 000, 1–16


components represent the leading edge of a cone and a morecentral component, with the trailing edge entirely absent. Inthis case, one expects an asymmetric PA swing (as seen) andalso increased circular polarization under the central com-ponent (as also seen). Generally, however, conal componentshave a flatter spectral index than core components and un-der this picture one might expect the leading componentto dominate at higher frequencies (since conal componentstend to dominate over core components at high frequencies)and this is not obviously the case. However, observationsover a wider frequency range (especially very low frequen-cies) might help solve this puzzle. One also has to explainwhy the trailing edge is entirely absent in these pulse pro-files, but this follows the trend noted by Lyne & Manchester(1988) that leading edge cones dominate over trailing edge.

The second possibility is that the two principal observedcomponents mark ‘classical’ double emission with the mag-netic pole crossing near the symmetrical centre of the pulseprofile. In this case, the asymmetry of the PA swing is thendue to A/R which has a greater effect in rapidly spinningpulsars than in slowly spinning ones. The question arises asto why the trailing component always dominates and hassignificant circular polarization. The single component pul-sars would then either be an extreme case where the lead-ing component is absent and only the trailing component ispresent, or a grazing cut through the outer edge of the emis-sion cone. The latter seems more plausible in light of the flatPA swings seen in the single component pulsars. This expla-nation goes against the trend shown by Lyne & Manchester(1988) where, in pulsar with partial cones, generally theleading component dominates. Taking all the data on youngpulsars into account, we also violate the correlation (in 9 outof 15 cases) between the handedness of circular polarizationand the sense of PA swing in conal pulsars found by Hanet al. (1998). On balance however, we favour this possibilityfor a number of reasons. First that PSR J1015−5719 clearlyshows a component between the two main components. Sec-ondly, both components generally have the same width, andthe symmetry point lies between the components. Finally,conal components generally have flat spectral indices as doyoung pulsars in general.

6.1 Emission Heights

If we assume the latter interpretation is the correct one, thenwe can use the results of the RVM fitting to derive the emis-sion height of the radiation. We do this by first deriving thelongitude of the midpoint of the pulsar profile using its in-herent symmetry, which we denote φs. Note that in Figures1−6 we have located φs at zero longitude as far as possi-ble. Then, we use the value of φ0 given in Table 2 to obtainthe emission height by computing δφ(PA) = φ0 − φs andattributing the difference to the effects of A/R. Applicationof equation 2 then leads to a determination of the emissionheight. These heights are listed in column 8 of Table 2; col-umn 9 lists the heights in terms of the fraction of the lightcylinder radius. From the emission heights we can computethe cone half-opening angle at the last open field line, ρo, viaequation 4. These values are shown in column 10 of Table 2.The values of ρo are distributed between 10 and 30 degrees.

The results show that the emission height for all thesepulsars lies between 5 and 15 per cent of the light cylinder.

This is significantly higher than the 1-2 per cent heightsfound for older pulsars (Gupta & Gangadhara 2003). Wecaution however against interpreting these values too liter-ally. In recent years evidence has built up that the emissionheight at a given frequency is not constant across the beambut is higher at the edges of the cone and lower in the middle(Gupta & Gangadhara 2003). If this is the case, A/R affectsthe core and cone by different amounts and the symmetri-cal midpoint of the profile will not necessarily correspond tothe core location and the emission heights computed throughA/R will be underestimates. Finally, other effects such as re-fraction effects (Petrova 2000) and magnetic field sweepbackcan distort this picture still further (Arendt & Eilek 1998;Dyks & Harding 2004).

We note there is a strong inconsistency in the inter-pretation of the data on PSR J1015−5719. The RVM fitshows the pulsar is nearly an orthogonal rotator; if this isthe case then the very large pulse width naturally impliesthe emission must come from high in the magnetosphere.On the other hand, the value of the emission height derivedfrom the location of φ0 is rather small. There are two pos-sible sources of error or confusion. The RVM fit might beincorrect and the pulsar is really an aligned, rather than or-thogonal rotator. This is possible, witness for example thecontinuing discussion on whether PSR B0950+08 is alignedor orthogonal (Everett &Weisberg 2001), but seems unlikelygiven the goodness of fit. Secondly, assigning the offset in φ0as arising due to A/R is in error due to the reasons we sug-gested above.

However, there exists a further appealing solution forPSR J1015−5719. This comes from recognising that themeasured A/R yields the emission height of the core emis-sion and may not reflect the emission height of the conalemission if this occurs significantly higher in the magneto-sphere than the core emission (Gupta & Gangadhara 2003).To test this, we have therefore decomposed the profile ofPSR J1015−5719 into 7 Gaussians, 3 for the leading edge,a central component and 3 for the trailing edge. The loca-tion of the centroid of these Gaussians occurs at longitudes−72◦, −52◦, −36◦, 0◦, 31◦, 44◦ and 56◦(cf Figure 6). Imme-diately one can see that the locations are asymmetric, withthe trailing components closer to the core component thanthe leading components, exactly as expected in the pictureof Gangadhara & Gupta (2001). We can use this asymmetry,in conjunction with the derived emission height of ∼400 kmfor the core component to derive the heights of the 3 conalcomponents (see equation 5). These are (from inner to outer)550 km, 650 km and 850 km respectively. Using the moreexact formalism of Gangadhara (2005) which takes into ac-count the geometrical angles the height of the outermostcone increases to 925 km. This helps solves the problem ofhaving an orthogonal rotator with a seemingly small emis-sion height and a wide profile. Indeed the emission height ofthe core is small, but the outer cones are at a significantlyhigher height, yielding a wide profile.

6.2 A Generic Young Pulsar Beam Model

We have concluded that young pulsar beams generally con-sist of a hollow cone, with little or no emission from thecenter. The double profiles arise from cuts close to the mag-netic pole, the single profiles are grazing trajectories along

c© 2006 RAS, MNRAS 000, 1–16


Figure 8. Simulation of a pulse profile and PA swing from oursimple model. Parameters are α=30◦, β=5◦, and we assume apulse period of 100 ms with an emission height of 250 km. Theleft panel shows the emission pattern with the line of sight tra-verse superposed. The top right panel shows the resultant pulseprofile with the bottom right panel showing the PA swing, whichis delayed relative to the total intensity profile by an amountconsistent with the emission height and the effects of A/R.

the cone. We quantify this by adopting a canonical youngpulsar with a spin period of 0.1 s. The light cylinder radiusis at ∼5000 km with the emission zone located at ∼250 km,consistent with the results in Table 2. These parameters nat-urally lead to a beam half-opening angle of ∼20◦. A simplebeam model can then be created in which ρ=20◦ and thereis a single hollow cone of emission located at the beam edge.The conal emission zone is relatively wide, has a Gaussianamplitude distribution and takes up∼0.2 of the beam. Thereis no core emission. We then choose random values of α andβ and compute the line of sight relative to the beam in orderto produce a simulated pulse profile. At the same time, wecan construct a PA swing based on the geometric angles,assuming the RVM is correct. The PA swing is then delayedin longitude with respect to the profile by an amount consis-tent with the emission heights and the effects of A/R. Fig 8shows an example. We find from the simulations that ∼50per cent of pulsars are potentially detectable (i.e. we havesome sightline across the open field lines); of these ∼45 percent have double profiles and ∼55 per cent have single pro-files. Most of the double component pulsars have widths of∼40◦, with a small minority having much wider beams. Forthe single components, there is a ratio of 2:1 between nar-row and wide profiles. The results from this simple simula-tion are very similar to the observed results. In the observedsample, there is roughly an equal split between profiles withsingle components and those with two components. Most ofthe double component profiles have similar widths, with theoccasional wide double. In contrast the single componentpulsars show a larger variety in pulse widths as seen in thesimulation.

Figure 9. Radio continuum image of SNR G308.8−0.1 (Caswellet al. 1992). The plus marks the position of PSR J1341−6220. Thesolid line denotes PA0 from our measurements. The dashed line isorthogonal to PA0 and is our preferred proper motion direction.

7 DERIVED PROPER MOTION DIRECTIONSAND THEIR IMPLICATIONS

As discussed in the introduction, we can potentially deter-mine the direction of motion of the pulsars through knowl-edge of the position angle of the rotation axis under the as-sumption that they are aligned with the velocity direction.Johnston et al. (2005) pointed out that, even if the vectorsare aligned, the polarization PA at φ0 could be perpendicu-lar to the rotation axis if the pulsar emits in the orthogonalmode. We have seen from the discussion of the pulse pro-files, it is very difficult to correctly assess the longitude ofthe magnetic pole in these young pulsars. We can only re-ally be confident in the cases where the RVM fit returnsan accurate value of φ and hence PA0 as listed in Table 2above. This is only the case for 7 pulsars in our sample, andone of these (PSR J1119−6127) has no polarization at thelocation of φ0. We describe the other six pulsars here. Thevalue of PA0 has an inherent 180

◦ ambiguity because of thedefinition of position angle of linear polarization. Also, it ispossible that PA0 can be 90

◦ different from the true valueof the rotation axis position angle, if the pulsar is emittingin the orthogonal mode (for full details see Johnston et al.2005). Hence in what follows, we assume that either PA0 orPA0±90

◦ will be parallel to the velocity vector of the pulsar.This implicitly assumes that velocity vector is aligned withthe rotation axis as found by Johnston et al. (2005) but wecaution that this might not be a general rule.PSRs J0729−1448 and J0940−5048: These two pulsarshave no apparent associations with high energy emission orSNRs. The value of PA0 and hence the direction of the ro-tation axis is parallel or perpendicular to −82.2 and −85.4degrees respectively.PSR J1105−6107: In this pulsar, we showed in Section2 that one might expect the velocity vector to point at135◦ if the pulsar was associated with SNR G290.1−0.8. Thevalue of PA0 that we derive from the RVM fit, −2.2

◦, wouldthen seem to imply that the pulsar did not originate fromSNR G290.1−0.8. The pulsar has a relatively large charac-

c© 2006 RAS, MNRAS 000, 1–16


Figure 10. Radio continuum image of SNR G320.4−1.2.PSR J1513−5908 lies at the intersection of the two cones whichrepresent a possible outflow geometry based on the continuummorphology (Gaensler et al. 1999). The solid line denotes PA0from our measurements. The dashed line is orthogonal to PA0and is our preferred proper motion direction.

teristic age and it may be that its parent SNR has simplydissipated.PSR J1341−6220: The main axis of symmetry of the pe-culiar SNR in which PSR J1341−6220 is embedded has aposition angle of −20◦, which is ∼90◦ offset from our mea-sured PA0 of 72.5

◦. It seems likely therefore that the pulsaris moving along this symmetry axis, the rotation axis alsopoints in this direction and it emits in the orthogonal mode.Figure 9 shows the SNR/PSR association with our derivedvalue of PA0 and the preferred proper motion direction.PSR J1420−6048: The pulsar lies in a highly complexregion of the Galactic plane as detailed above. A propermotion in virtually any direction would intersect somethingof interest! It so happens that the perpendicular to ourPA0 of −61

◦ points at the centre of the ‘Kookaburra’ com-plex. Is PSR J1420−6048 a pulsar moving at ∼400 km s−1

which originated in the centre of the ‘Kookaburra’, punc-tured through the shell and is currently creating a wind bub-ble similar to the picture seen in the ‘Duck’ system (Gaensler& Frail 2000; Thorsett, Brisken & Goss 2002)?PSR J1513−5908: The PA of the symmetry axis of thePWN in which the pulsar is embedded is 150◦±5◦. Gaensleret al. (2002) infer that the rotation axis of the pulsar mustalso have this PA to conform with their model. Our polar-ization measurements yield PA0 of 63

◦, which differs by 87◦

from the Gaensler et al. (2002) conjecture. We therefore sup-port the idea that the PWN symmetry axis is indeed alignedwith the pulsar’s rotation axis and that the pulsar emits inthe orthogonal mode as in the Vela pulsar (Johnston et al.2005) and the above case of PSR J1341−6220. If the correla-tion between the spin axis and the velocity vector in pulsarsis also correct then we predict the pulsar’s proper motionwill also lie along this axis. Figure 10 shows the SNR/PSRassociation with our derived value of PA0 and our proposedproper motion direction.

8 CONCLUSIONS

We have observed 14 young pulsars and obtained calibratedpolarization profiles. The profiles of young pulsars are gen-erally simple and fall into two categories. In the first, theprofiles consist of two, highly polarized components with aflat PA swing across the first component and a steep swingacross the second. The second component is nearly alwaysbrighter than the first component. The correlation pointedout by Han et al. (1998) between the sign of circular po-larization and the sense of swing of the PA appears not toapply in these double pulsars. The second category consistsof a simple, broad single component with high polarizationand a shallow swing of position angle.

We interpret these results as showing that the emissionbeams of young pulsars consist simply of a single emissioncone located near the last open field lines. Core emissionis absent or rather weak. Emission arises between 1 and 10percent of the light cylinder radius which is significantlyhigher than the emission zone in older pulsars. This sim-ple model explains many of the features of the observations.Two questions remain: why are the trailing components al-ways brighter and why do they generally have more circularpolarization?

We predict that PSRs J1341−6220 and J1513−5908have aligned rotation and velocity axes and it would be use-ful to measure the proper motion of these pulsars as verifi-cation.

ACKNOWLEDGMENTS

The Australia Telescope is funded by the Commonwealthof Australia for operation as a National Facility man-aged by the CSIRO. JMW was supported by Grant AST0406832 from the U.S. National Science Foundation. Wethank M. Kramer for the RVM fitting routines. We thankthe referee, R. T. Gangadhara, for his careful reading of themanuscript.

REFERENCES

Arendt P. N., Eilek J. A., 1998, ApJ, Submitted

Bilitza D., 2003, Advances in Space Research, 31, 757Blaskiewicz M., Cordes J. M., Wasserman I., 1991, ApJ, 370, 643Brazier K. T. S., Becker W., 1997, MNRAS, 284, 335Camilo F., Kaspi V. M., Lyne A. G., Manchester R. N., Bell J. F.,

D’Amico N., McKay N. P. F., Crawford F., 2000, ApJ, 541,367

Camilo F. et al., 2001, ApJ, 557, L51Case G., Bhattacharya D., 1999, ApJ, 521, 246Caswell J. L., Kesteven M. J., Stewart R. T., Milne D. K.,

Haynes R. H., 1992, ApJ, 399, L151Crawford F., Keim N. C., 2003, ApJ, 590, 1020Crawford F., Manchester R. N., Kaspi V. M., 2001, AJ, 122, 2001Crawford F., Gaensler B. M., Kaspi V. M., Manchester R. N.,

Camilo F., Lyne A. G., Pivovaroff M. J., 2001, ApJ, 554, 152D’Amico N. et al., 2001, ApJ, 552, L45

Doherty M., Johnston S., Green A. J., Roberts M. S. E., Ro-mani R. W., Gaensler B. M., Crawford F., 2003, MNRAS,339, 1048

Dyks J., Harding A. K., 2004, ApJ, 614, 869Dyks J., Rudak B., Harding A. K., 2004, ApJ, 607, 939Everett J. E., Weisberg J. M., 2001, ApJ, 553, 341

c© 2006 RAS, MNRAS 000, 1–16


Gaensler B. M., Frail D. A., 2000, Nat, 406, 158

Gaensler B. M., Brazier K. T. S., Manchester R. N., Johnston S.,Green A. J., 1999, MNRAS, 305, 724

Gaensler B. M., Arons J., Kaspi V. M., Pivovaroff M. J.,Kawai N., Tamura K., 2002, ApJ, 569, 878

Gangadhara R. T., Gupta Y., 2001, ApJ, 555, 31Gangadhara R. T., 2005, ApJ, 628, 923Gil J. A., Gronkowski P., Rudnicki W., 1984, AA, 132, 312Gonzalez M. E., Kaspi V. M., Camilo F., Gaensler B. M., Pivo-

varoff M. J., 2005, ApJ, 630, 489Gupta Y., Gangadhara R. T., 2003, ApJ, 584, 418Han J. L., Manchester R. N., Xu R. X., Qiao G. J., 1998, MNRAS,

300, 373Huguenin G. R., Manchester R. N., Taylor J. H., 1971, ApJ, 169,

97Johnston S., Hobbs G., Vigeland S., Kramer M., Weisberg J. M.,

Lyne A. G., 2005, MNRAS, 364, 1397Karastergiou A., Johnston S., Manchester R. N., 2005, MNRAS,

359, 481Kaspi V. M., Manchester R. N., Johnston S., Lyne A. G.,

D’Amico N., 1992, ApJ, 399, L155Kaspi V. M., Bailes M., Manchester R. N., Stappers B. W.,

Sandhu J. S., Navarro J., D’Amico N., 1997, ApJ, 485, 820Kramer M. et al., 2003, MNRAS, 342, 1299Lyne A. G., Manchester R. N., 1988, MNRAS, 234, 477Manchester R. N., Johnston S., 1995, ApJ, 441, L65Manchester R. N., 1996, in Johnston S., Walker M. A., Bailes M.,

eds, Pulsars: Problems and Progress, IAU Colloquium 160.Astronomical Society of the Pacific, San Francisco, p. 193

Manchester R. N., D’Amico N., Tuohy I. R., 1985, MNRAS, 212,975

Manchester R. N. et al., 2001, MNRAS, 328, 17Manchester R. N., Tuohy I. R., D’Amico N., 1982, ApJ, 262, L31Mitra D., Li X. H., 2004, AA, 421, 215Petrova S. A., 2000, AA, 360, 592Pivovaroff M. J., Kaspi V. M., Camilo F., Gaensler B. M., Craw-

ford F., 2001, ApJ, 554, 161Qiao G. J., Manchester R. N., Lyne A. G., Gould D. M., 1995,

MNRAS, 274, 572Radhakrishnan V., Cooke D. J., 1969, Astrophys. Lett., 3, 225Rankin J. M., 1990, ApJ, 352, 247Roberts M. S. E., Romani R. W., Johnston S., Green A. J., 1999,

ApJ, 515, 712Roberts M. S. E., Romani R. W., Johnston S., 2001, ApJ, 561,

L187Romani R. W., Yadigaroglu I.-A., 1995, ApJ, 438, 314Seward F. D., Harnden Jr. F. R., 1982, ApJ, 256, L45

Thorsett S. E., Brisken W. F., Goss W. M., 2002, ApJ, 573, L111Torres D. F., Butt Y. M., Camilo F., 2001, ApJ, 560, L155von Hoensbroech A., Kijak J., Krawczyk A., 1998, AA, 334, 571Yatsu Y., Kawai N., Kataoka J., Kotani T., Tamura K.,

Brinkmann W., 2005, ApJ, 631, 312You X.-P., Han J.-L., 2005, Chinese Journal of Astronomy & As-

trophysics, In Press

c© 2006 RAS, MNRAS 000, 1–16

arxiv:astro-ph/0603037v1 2 mar 2006se direction (135 ) but this has not yet been measured. there is...

Documents